EMH means that well informed markets make sound decisions on the pricing of assets. This means that based on the current information available, markets are excellent at understanding the probability-weighted value of an asset. It doesn't mean that the outcome ends up being right, it means that it's fairly priced based on the information at hand. Nothing guaranteeing Pareto-optimal outcomes from that.
Furthermore, an investor giving 1.2M to a company is not an efficient market under any circumstance.
Here is one well stated version of the EMH:
"The Efficient Market Hypothesis (EMH) essentially says that all known information about investment securities, such as stocks, is already factored into the prices of those securities. Therefore no amount of analysis can give an investor an edge over other investors. EMH does not require that investors be rational; it says that individual investors will act randomly but, as a whole, the market is always "right." In simple terms, "efficient" implies "normal." For example, an unusual reaction to unusual information is normal."[1]
[1] http://mutualfunds.about.com/od/mutualfundglossary/a/Efficie...
Well, there's a subtlety there.
It's not that "free markets work best". It's that the efficient market hypothesis claims that a market with perfect information dissemination consisting of uniformly rational agents will deliver Pareto-optimal resource allocations.
There's a lot to disagree with in that sentence; market agents collude, people are demonstrably irrational/rely on cognitive heuristics and just because something is Pareto efficient doesn't mean it's fair or equitable.
All three are valid critiques, and the case in point in TFA - the market allocated money to Yo, but that doesn't mean a priori that funding Yo is a worthy and moral choice.
So, if we have markets that routinely deliver inequitable outcomes it's perfectly reasonable to ask - why is this happening?